We used the Prokka tool (20) to annotate 2,493 genomes belonging to ST101, ST307, and ST512 (Appendix 3, Table). We used Roary (24) to generate core-genome alignments, using MAFFT (25), accordingly, to the ST and IQ-TREE 2 (26) to construct phylogenetic trees with 1,000 ultrafast bootstrap iterations.
Our initial aim with this study was to explain the different levels of FDC resistance in 4 ST512 K. pneumoniae clinical isolates (PL1, PL2, PL3, and PL4) from 1 patient during 1 month of hospitalization. PL3 was resistant to FDC (MIC 4 mg/L), whereas FDC MICs PL1, PL2, and PL4 were below the breakpoint value (0.5-2.0 mg/L; Appendix 1, Table 1). The PL1-4 genomes were closely related at the chromosomal level (6-22 single-nucleotide polymorphisms and indels on the chromosome; Appendix 4, Table) but showed different bla genes and plasmid content. All PL1-4 K. pneumoniae strains carried different pKpQIL variants plus an IncX3 plasmid and a small ColRNAI plasmid. The pKpQIL-PL1 plasmid harbored copies of the bla gene, both pKpQIL-PL2 and pKpQIL-PL4 carried 1 copy of bla and 1 copy of bla, whereas pKpQIL-PL3 carried 2 copies of bla (Appendix 1, Figure 2). Isolates PL2 and PL3 were also enriched with the pKPN plasmid, which was absent in PL1 and PL4. In PL2, pKPN was fused with pKpQIL-PL2 in the tnpR-FIIK integration site, forming a 263,486-bp plasmid. The hybrid pKPN-pKpQIL-PL2 plasmid was not transferable by transformation or conjugation and could not be further studied. In PL3, the stand-alone pKPN plasmid had acquired the fec gene cluster encoding for a FEC system. The fec gene cluster was unique to the pKPN plasmid in PL3 and was absent in PL1, PL2, and PL4. The fec genes mapped (alongside an ABC glutathione transporter, the lacZ, lacY, and lacI genes) between 2 IS4321 elements positioned between the tnpR gene and the FIIK replicon (Appendix 1, Figure 3).
Acquisition of fec genes was suspected to correlate with increased FDC MICs of PL3. We then measured FDC MICs for all KPC-producing K. pneumoniae clinical strains in our collection isolated since 2018 with a completely sequenced genome, in search for fec, other siderophore receptors, and porin gene sequences in their genomes (Appendix 1, Table 1).
FDC MICs were 0.25-32 mg/L. The lowest (0.25 mg/L) was measured in a strain producing KPC-3 (strain 3), encoding the yersiniabactin siderophore-dependent iron uptake system, the wild-type CirA and Fiu siderophore receptors, and a wild-type OmpK36 porin (27). The highest FDC MIC (32 mg/L) was for a strain producing VIM (Verona integron-encoded metallo-β-lactamase) and KPC carbapenemases (strain 0296), characterized by a nonsense mutation in the gene encoding the siderophore receptor CirA (E133X) (28).
Presence of CZA-resistant variant KPC-31, KPC-70, or KPC-68 was associated with high MICs (4 mg/L). Higher MICs (1-2 mg/L) for FDC were obtained in E. coli Top-10 transformed with the bla, bla, or bla genes, respectively cloned in the pTopo-KanR vector (Appendix 1, Table 1).
In addition to the role of KPC variants in determining FDC resistance levels, we noticed that K. pneumoniae exhibiting higher FDC MICs were positive for the FEC system (29), carried by the pKPN plasmid (30). The fec gene cluster was identified in 13/35 isolates from the collection. Eleven KPC-31-producing strains belonging to different STs showed FDC MICs of 1-2 mg/L, but 2 KPC-31 strains carrying the fec gene cluster reached a MIC of 4.0 mg/L (Appendix 1, Table 1). We hypothesized that the plasmid-borne FEC system could reduce the susceptibility to FDC in K. pneumoniae clinical isolates.